-
-
Save mrdoob/1325393 to your computer and use it in GitHub Desktop.
ffmpeg -f x11grab -b 1M -bt 2M -r 30 -s 512x512 -i :0.0+1,53 -an kinect.webm |
@chriskilding :
I think you refers about the 16 bits requiered when working with grayscale as usually for kinect: 1 <=> 1mm, thus 16bits means max precision of 2^16-1=65536=65 meters. Max distance of Kinect is <10meters
But here the video captured has colors (https://threejs.org/examples/textures/kinect.webm) and on a computer usually the image displayed is 8bits per channel, thus rgb <=> 2^(8*3) = 16777216, so no information is lost (but probably some frame will be missing or duplicated)
Also, if working in browser, we cannot rely on browser PNG/JPG decoder to view the 16 bits grayscale image, as it only handle 8 bits per channel rgb thus 8 bits grayscale. In that case you will have read the image as binary and decode it youself to get the value of each pixel
Hi, could you confirm whether the [OpenNI viewer -> ffmpeg's x11grab -> WebM/VP8 codec] combination captures all 16 bits/channel necessary to encode the Kinect depth map fully, or is it downsampled / truncated to 8 bits?